South KoreaPublic sector2026 – ongoingInstitute for Basic Science
Cross-material catalyst discovery with a crossbreeding neural network
Institute for Basic Science researchers tested catalysts predicted by a cross-material neural network
Problem
Catalyst research usually stayed within individual material families. This limited the use of experimental findings from other catalyst types.
Approach
IBS-affiliated researchers developed a crossbreeding neural network using shared chemical descriptors from separate experimental datasets. They tested predictions against synthesized catalysts in a previously untrained material class. The model predicted performance trends, including a multimetallic catalyst that exceeded previous candidates.
Results
No published figures
Takeaway
The researchers checked model predictions against physical experiments. They did not quantify research time or cost savings.
Categories
- Industry
- Public sector
- Business function
- IT and R&D
- Innovation type
- Process innovation
- AI technology
- Predictive analytics and ML
- Adoption
- Built in-house
- Evidence
- Independent study
Sources
Source summaries
The researchers described a neural network connecting different catalyst families. Experimental synthesis supported its predicted activity trends, including a catalyst outperforming earlier candidates.
Full details are in the original sources.